series.ts 3.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103
  1. 'use strict';
  2. import cheerio = require('cheerio');
  3. import episode from './episode';
  4. import fs = require('fs');
  5. import request = require('./request');
  6. import path = require('path');
  7. import url = require('url');
  8. var persistent = '.crpersistent';
  9. /**
  10. * Streams the series to disk.
  11. */
  12. export default function(config: IConfig, address: string, done: (err: Error) => void) {
  13. var persistentPath = path.join(config.output || process.cwd(), persistent);
  14. fs.readFile(persistentPath, 'utf8', (err, contents) => {
  15. var cache = config.cache ? {} : JSON.parse(contents || '{}');
  16. page(config, address, (err, page) => {
  17. if (err) return done(err);
  18. var i = 0;
  19. (function next() {
  20. if (i >= page.episodes.length) return done(null);
  21. download(cache, config, address, page.episodes[i], (err, ignored) => {
  22. if (err) return done(err);
  23. if ((ignored == false) || (ignored == undefined))
  24. {
  25. var newCache = JSON.stringify(cache, null, ' ');
  26. fs.writeFile(persistentPath, newCache, err => {
  27. if (err) return done(err);
  28. i += 1;
  29. next();
  30. });
  31. }
  32. else
  33. {
  34. i += 1;
  35. next();
  36. }
  37. });
  38. })();
  39. });
  40. });
  41. }
  42. /**
  43. * Downloads the episode.
  44. */
  45. function download(cache: {[address: string]: number},
  46. config: IConfig,
  47. baseAddress: string,
  48. item: ISeriesEpisode,
  49. done: (err: Error, ign: boolean) => void) {
  50. if (!filter(config, item)) return done(null, false);
  51. var address = url.resolve(baseAddress, item.address);
  52. if (cache[address]) return done(null, false);
  53. episode(config, address, (err, ignored) => {
  54. if (err) return done(err, false);
  55. cache[address] = Date.now();
  56. done(null, ignored);
  57. });
  58. }
  59. /**
  60. * Filters the item based on the configuration.
  61. */
  62. function filter(config: IConfig, item: ISeriesEpisode) {
  63. // Filter on chapter.
  64. var episodeFilter = config.episode;
  65. if (episodeFilter > 0 && parseInt(item.episode, 10) <= episodeFilter) return false;
  66. if (episodeFilter < 0 && parseInt(item.episode, 10) >= -episodeFilter) return false;
  67. // Filter on volume.
  68. var volumeFilter = config.volume;
  69. if (volumeFilter > 0 && item.volume <= volumeFilter) return false;
  70. if (volumeFilter < 0 && item.volume >= -volumeFilter) return false;
  71. return true;
  72. }
  73. /**
  74. * Requests the page and scrapes the episodes and series.
  75. */
  76. function page(config: IConfig, address: string, done: (err: Error, result?: ISeries) => void) {
  77. request.get(config, address, (err, result) => {
  78. if (err) return done(err);
  79. var $ = cheerio.load(result);
  80. var title = $('span[itemprop=name]').text();
  81. if (!title) return done(new Error('Invalid page.(' + address + ')'));
  82. var episodes: ISeriesEpisode[] = [];
  83. $('.episode').each((i, el) => {
  84. if ($(el).children('img[src*=coming_soon]').length) return;
  85. var volume = /([0-9]+)\s*$/.exec($(el).closest('ul').prev('a').text());
  86. var regexp = /Episode\s+((PV )?[S0-9][P0-9.]*[a-fA-F]?)\s*$/i;
  87. var episode = regexp.exec($(el).children('.series-title').text());
  88. var address = $(el).attr('href');
  89. if (!address || !episode) return;
  90. episodes.push({
  91. address: address,
  92. episode: episode[1],
  93. volume: volume ? parseInt(volume[0], 10) : 1
  94. });
  95. });
  96. done(null, {episodes: episodes.reverse(), series: title});
  97. });
  98. }